About Official Ampersand AI SDKs
Ampersand’s official SDK and MCP server help AI application builders manage integrations with customers’ B2B SaaS tools. The SDK includes modules for Vercel AI SDK and Mastra, with examples for agent frameworks.
Key features
- 01Manage customer SaaS integrations
- 02Vercel AI SDK module
- 03Mastra module
- 04Examples for agent frameworks
Use cases
- 01Create or update integration records in agents
- 02Check Salesforce connections
- 03Use Ampersand integration tools through MCP
MCP support
The official MCP server exposes tools from the Ampersand AI SDK to MCP clients and connects agents to Ampersand platform connectors. It can be configured over SSE or stdio.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEOfficial Ampersand AI SDKs
This repo contains the official Ampersand AI SDK and the offical MCP server for Ampersand.
- AI SDK - Official Ampersand AI SDK that exposes tools for your AI agents to manage and interact with integrations with your customer's SaaS tools.
- MCP server - Official Ampersand MCP server that exposes the tools from the Ampersand AI SDK.
The examples directory contains examples of how to use the AI SDK with popular agent frameworks.
Ampersand is a platform for AI builders to add B2B SaaS integrations to their agentic apps. To learn more, visit our website.
Ampersand AI SDK
Installation
npm install @amp-labs/ai
# or
yarn add @amp-labs/ai
# or
pnpm add @amp-labs/ai
Usage
The SDK provides several modules that can be used depending on your framework preference:
Using with Vercel AI SDK
import { createRecord, updateRecord } from "@amp-labs/ai/aisdk";
// Use in your AI agent configuration
const tools = [createRecord, updateRecord];
Using with Mastra
import { createRecord, updateRecord } from "@amp-labs/ai/mastra";
// Use in your Mastra workflow
const tools = [createRecord, updateRecord];
Testing
The test directory contains end-to-end tests for the AI SDK.
Running the Simple Test
Prerequisites:
- OpenAI API key (required for AI model)
- Ampersand API credentials
-
Configure environment variables in
test/.env:OPENAI_API_KEY=your_openai_api_key_here AMPERSAND_API_KEY=your_ampersand_api_key_here AMPERSAND_PROJECT_ID=your_project_id_here AMPERSAND_GROUP_REF=your_group_ref_here -
Run the test:
# From the root directory pnpm --filter ai-e2e-test test:simple # Or from the test directory cd test pnpm test:simple
The simple test demonstrates how to use the checkConnection tool with the Vercel AI SDK to verify Salesforce connections.
Testing Local Changes
The test uses your local SDK build via pnpm workspace linking. To test modifications to the SDK:
-
Make changes to SDK source files in
sdk/lib/ -
Rebuild the SDK:
pnpm --filter @amp-labs/ai build -
Re-run the test to verify your changes:
pnpm --filter ai-e2e-test test:simple
The test will use the freshly built SDK from sdk/dist/, allowing you to iterate quickly on SDK changes.
Ampersand MCP Server
Connect your agents to the 150+ connectors we offer at Ampersand via this multi-tenant MCP server. We expose the primitives we offer on the Ampersand platform as native tools here.
Connect to the remote MCP server
Add the following in your mcp.json in Cursor IDE or claude_desktop_config.json when using Claude desktop.
When running the server in SSE mode
If your MCP client supports headers:
{
"mcpServers": {
"@amp-labs/mcp-server": {
"url": "https://mcp.withampersand.com/v1/sse?project=<AMPERSAND_PROJECT_ID>&integrationName=<AMPERSAND_INTEGRATION_NAME>&groupRef=<AMPERSAND_GROUP_REF>",
"headers": {
"x-api-key": "<AMPERSAND_API_KEY>"
}
}
}
}
If your MCP client does not support headers, you can pass the API key in the query param:
{
"mcpServers": {
"@amp-labs/mcp-server": {
"url": "https://mcp.withampersand.com/v1/sse?apiKey=<AMPERSAND_API_KEY>&project=<AMPERSAND_PROJECT_ID>&integrationName=<AMPERSAND_INTEGRATION_NAME>&groupRef=<AMPERSAND_GROUP_REF>"
}
}
}
When running the server in stdio mode
{
"mcpServers": {
"@amp-labs/mcp-server": {
"command": "npx",
"args": [
"-y",
"@amp-labs/mcp-server@latest",
"--transport",
"stdio",
"--project",
"<AMPERSAND_PROJECT_ID>",
"--integrationName",
"<AMPERSAND_INTEGRATION_NAME>",
"--groupRef",
"<AMPERSAND_GROUP_REF>"
],
"env": {
"AMPERSAND_API_KEY": "<AMPERSAND_API_KEY>"
}
}
}
}
Connect to the local MCP server
Please refer to the mcp-server README for more information.
License
This project is licensed under the MIT license.
See LICENSE for more information.
Contributing
We welcome contributions! Please refer to the CONTRIBUTING.md file for more information.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.